Program yuan, AI, algorithms, face recognition, Ali Security Turing Lab… Any one of these labels is highly topical, but what would it be like for a woman to have them all?
Before I saw Yu Kun, my mind was filled with all kinds of imagination, cute girl? Royal elder sister? Still? Although Yu Kun speaks a little fast, she is crisp and logical, so “capable” is the deepest impression she leaves on me.
The AI algorithm program of Ali Security Turing Lab is Yuan Kun
After years of training in Alibaba, Yu Kun has become a top expert in the field of face core algorithm in China. As a core member of Ali Security Turing Lab, I have developed the industry’s top level of face recognition technology and algorithm. Whether 1:1 face comparison or 1:N face search algorithm, the effect is in an advantageous position in the industry. At present, the online face comparison core algorithm model is trained and developed by Yu Kun and his friends.
Not only that, Yu Kun also excellently close to the business scene, the perfect implementation of the technical achievements. She led the face comparison algorithm of new retail stores (such as Future Meijia, Intime Department Store, Hema, etc.) and other 1: N offline businesses (including access control in the park, pickup cabinet of Bird Express, etc.); At the same time, it also cooperated with the research and development of important business algorithms such as 1:1 real person authentication (hand cleaning face login, etc.) and 1: N sensitive face detection (internal and green web).
Interacting directly with users is’ exciting ‘
In 2014, Yu kun came into contact with Alibaba and was fascinated.
“I think real person authentication and content control is very promising and can interact directly with the end user.” “It’s very attractive and rewarding for those who are in the technology business,” yu said.
Yu Kun believes, such as real person certification, if the technical level is not done, it will lead to a decline in experience. But when the customer’s process fails for technical or business reasons, the user gives the most direct feedback. For example, scan face login, users’ response is the most intuitive, the limit time that users can accept is 2 seconds, which puts forward high requirements for technology.
“Direct feedback from users is one of the most rewarding experiences for technical people. “Yu Kun stressed.
Because the algorithm is at the back end, yu kun found the interaction with the foreground user to be a “very exciting” experience.
In Yu kun’s eyes, alibaba’s big platform is really attractive. Human authentication is huge, hundreds of millions of people use it. For most technical people, this is not an easy opportunity to get.
It’s a new experience at Alibaba
Since 2007, Yu Kun has been doing research and development of face recognition technology.
Technically, Yu knows his way around, but that doesn’t mean the road ahead is smooth. “For me, it was actually quite challenging because it required a huge character shift.” Yu Kun said.
Yu Kun introduced that she had been doing face technology research and development in traditional enterprises, but to Alibaba such a head Internet company, need to reform themselves. Including corporate culture, working process, etc., these are all new experiences for Kun.
“The general feeling was that Ali’s pace was very fast and he was a little uncomfortable at first.” Yu Kun said. She felt that she was trotting forward every day, facing complicated business and connecting with many teams. Most of the day is spent in meetings, and it’s mostly after dinner that I have my own time, when I sit down to do something. “Basically, you’re either in meetings or on the way to meetings during the day. “Yu Kun laughed.
All things are difficult before they are easy. However, through continuous communication and run-in, Yu kun and his friends have more and more tacit understanding, communication is more and more smooth, and more and more understanding of the business, the project is also “handy”. Two years later, Yu Kun completely adapted to the fast pace and new environment. It’s natural to think, “This is how it should be.”
“Now, if it’s a slow pace, I’m not used to it. I feel like I’m wasting my time and my life.” “That’s how work is supposed to work,” Yu said.
Yu Kun told me that she felt she had opened a new page and experienced a different life when she joined Alibaba.
The programmer is not a unique creature
The research direction of Ali Safe Turing experiment is very stable, which focuses on computer vision, natural language processing, machine learning and deep learning. Its focus on business is also very stable, that is, to provide technical support for core business lines such as Mobile, Tmall, Alipay, Xianyu, Youku Tudou, UC, Ali Mama, and export AI capabilities. As a result, the team is very stable.
Ali Security Turing Lab part of the technical students
In recent years, Yu Kun has been working at Ali Security Turing Laboratory. According to her introduction, the whole team has 5 program yuan, 3 Base in Hangzhou, 2 Base in Beijing. “The atmosphere of our team is very good. When we are together, we are ‘buddies’. We share good ideas with each other, like a family.” “Everyone is working in the same direction and toward the same goal,” Yu said.
Yu Kun believes that under the big goal of the team, only the division of labor is different, and the small partners more complement each other and cooperate back-to-back. “I leave the matter to you, and the matter ends here. It’s that simple.” Yu Kun said.
Yu Kun said with a smile that she kept her hair long just to retain her girl characteristics, but she never treated herself as a special existence to enjoy special treatment. On the contrary, she thinks she has been in the business longer and should take on more responsibilities.
Let your ideas flow
The Leader is very important to the stability of the team. This is Yu Kun’s opinion.
In her eyes, Ultron, the head of Ali Security’s Turing lab, is a man with ideas and a big picture in mind. Because of this, the atmosphere of the whole team is harmonious, and there is no division of ranks.
Yu Kun still remembers that one Mid-Autumn Festival, Ultron came to the company with hairy crabs and gave them to his team mates. “The Leader always thinks of his team members when he has anything to do, which is great!” Yu Kun said.
Compared with traditional enterprises and Internet companies, Yu Kun also talked about his understanding of the technical team. In her opinion, whether a team is good or not depends on whether the Leader has vision and foresight. In the technical team of traditional enterprises, the Leader takes the task and then gives it to the team members to implement, which is a top-down research and development mode. But in Alibaba, it is a bottom-up model, that is, the goal of the organization is there, and every team member can see it. What can everyone do for this goal? Team members are spontaneous, that is, alibaba has always emphasized self-organization and self-drive, and everyone is internally driven.
“The whole team is like that, the whole Ali is like that.” “Although there are KPIs, each team is thinking about what else they can do. Everyone is also thinking what can I do? The power is infinite.”
In Yu Kun’s opinion, top-down instructions are often just the thinking of the Leader, so the concrete results may not be good. But if it’s bottom-up, then everyone’s ideas can be brought to full play, everyone can contribute their intelligence to do useful work for the big goal, and it’s a kind of power that counts.
“Alibaba’s working model is that you have an idea and you run with it.” Yu Kun said.
Algorithmic people speak by numbers
In Alibaba’s Xixi industrial Park, face brushing devices have replaced the traditional access control by swiping cards. This is the masterpiece of Yu Kun and his team members. In fact, it was secured by Ali Security Turing LABS after a fierce internal technical battle.
The facial recognition technology developed by Yu Kun and his partner is applied to Alibaba’s access control system
“Algorithmic people rely on numbers to speak, who has the highest accuracy, who on.” Yu Kun told me that the face recognition system developed by Ali Security Turing Lab ranked first in terms of accuracy, thus winning the project of entrance guard face brushing in the park.
Ali Security Turing Laboratory has not only achieved good results internally, but also successfully won the security check project of an airport in the competition with external suppliers. Moreover, the laboratory has also served tens of thousands of enterprise customers through the two products of “real person certification” and “content security” on Ali Cloud.
“Our business depends on our technology, so our technology is not developed to win a competition, but to ensure the smooth application of the business and provide the best user experience.” Yu Kun told me that Ali Security Turing Lab often won awards in some competitions, but usually these were just a “sideline” for technical colleagues while working on business projects.
Yu Kun told me that the goal of Ali Security Turing Lab is to be the top algorithm team and maintain a leading position.
Use technology to change your life
Yu Kun gave me a brief history of the development of face recognition technology.
In the early 1990s, mobile phones were not popular, and the application of face recognition was still very preliminary, usually simple feature classification, such as geometric or gray features.
In 2003 and 2004, face recognition technology began to shift to the traditional features of the image field. At that time, it was more about the understanding of the image by engineers, which was very challenging. However, there are already some relatively preliminary commercial applications, such as some attendance machines.
By 2012, deep learning had exploded. With the development of this wave, face recognition is one of the most typical applications. At that point, face recognition is ready to take off, whether it’s data or algorithms.
In recent two or three years, face recognition is mainly in the cultivation of user habits, after all, there are still a lot of people are not able to accept, but the status quo is changing dramatically. For example, scan face payment, scan face transfer, etc., they have the convenience of any previous technology can not be matched.
Yu Kun told me that China is very ahead in facial recognition technology, and also in the cultivation of user habits. Every year, the country’s research investment in this field, such as the number of papers or patents, leads the world.
In fact, Yu Kun has been engaged in the research and development of facial recognition technology for 11 years, so many people will wonder if she is bored. Or do we need to go in a different direction? She had no such idea. She believes that face recognition is not a single point of technology, but a whole set of solutions, in addition to the solution itself, its underlying technology is also complex.
‘The face is constantly changing. Things like make-up, things like beauty treatments, things like aging, things like light exposure. These changes are far more complex than any other model. That’s a huge challenge for a machine.” “It is because of the complexity of face recognition that it has taken so long and so many people to study it,” Yu said.
Yu Kun thinks that face recognition is a very good point, there are many application scenarios and many commercial possibilities, so it can become the current hot spot.
“In recent years, the application of human face has become more and more widespread. In fact, we are not standing in the wind, but the whole industry is standing in the wind. “The crazy capital chase and the huge investment of companies all indicate that the market is promising.” “Yu kun said.
“I want to use technology to serve the public, really use technology to change life, use technology to eliminate evil and promote good, this is my dream, I believe it is also the dream of every technician.” Yu Kun said.
Original author:
China was
The original link